This study was carried out in accordance with the guidelines stipulated by the National Health Research Ethics Council of South Africa and approved by the University of Pretoria Animal Ethics Committee (AEC) with the approval number EC061-14.

The ATE, a water-soluble extract from the Acacia mearnsii tree bark was obtained from UCL Tannin Pty (Ltd), South Africa, and used as the active ingredient or core material in the current study. Gum arabic, maltodextrin (DE 16.5), native potato starch and tannic acid were obtained from Sigma Aldrich Inc. (St. Louis, MO, USA); the F57 fibre filter bags were purchased from ANKOM Inc. (Fairport, NY, USA). All chemicals and reagents were of analytical grade in purity.

The ATE, obtained from UCL Tannin Company Pty (Ltd), Dalton, South Africa, was extracted from the bark of black wattle (Acacia mearnsii) tree by steam distillation, and then concentrated into powdered form. The extract has a molecular weight that ranges from 500 to 3,000, with an average of 1,250 and it contains other non-tannins (including low molecular weight polyphenols, salts, sugars, and organic acids). The result of our laboratory analysis showed that the sample had total phenol and total tannin concentrations of 65.8% and 58.5% respectively (as tannic acid equivalent) according to the procedure of Makkar et al [13] and condensed tannin concentration of 30.5% (as leucocyanidin equivalent) according to the procedure of Porter et al [14]. Because of possible variation in extracts’ characteristics from the company, a single batch of extract, stored at 4°C was used for all preparations and analysis throughout this study.

The microparticles were prepared using a procedure similar to that of Zhang et al [8] with slight modifications. Briefly, wall materials were dissolved in water and homogenised. The ATE was added to the solution under continuous stirring and homogenization using an overhead homogenizer for 180 seconds (RW 20, Ika-Werke, Janke & Kunkel-Str. 10, Staufen, Germany). For the maltodextrin-Gum Arabic microparticles, the following parameters were fixed for all preparations: Gum-Arabic and Maltodextrin were added at the ratio of 40:60 (w/w); solute concentration in water was 1% (w/v) [9]. For the starch microparticles, native starch was suspended in ethanol:water (10:90 v/v), and heated under continuous mixing at 65°C to 70°C until a gelatinised paste was obtained. Gradual cooling was done with the addition of tannin extract under continuous homogenization as described by Fernandes et al [10]. The final mix was freeze-dried, and thereafter stored away in an airtight container. The ratio of wall to the core material in both preparations was varied from 85:15 to 65:35 in order to evaluate optimum loading conditions. The different preparation concentration of the wall material and ATE are designated as maltodextrin-gum Arabic tannin extract 25–35 (MG-TE_25–35) and starch-tannin extract 15–30 (S-TE_15–30).

The total phenol content of the ATE was evaluated using the Folin-Ciocateau colorimetric method and values were expressed as tannic acid equivalent [14] while the total condensed tannin content was evaluated using the butanol-HCl method and expressed as leucocyanidin equivalent [14]. For assaying the total phenol content, about 0.5 mL of extract was vortex mixed with 0.25 mL of the Folin-Ciocateau reagent and 1.25 mL of 20% sodium carbonate. The absorbance of the resultant solution was measured at 725 nm after 40 min and concentration estimated using a standard absorbance curve from tannic acid. Total polyphenolic contents were expressed as mg/g tannic acid equivalent. For the butanol-HCl method, 0.5 mL of sample containing extract, 3.0 mL butanol HCL (95:5 v/v) and 0.1 mL ferric reagent was added. Samples were vortexed and heated at 100°C for 60 min and absorbance read at 550 nm. A suitable blank containing the unheated sample was subtracted. For the quantification of EE, and rate of release of the active ingredient from microcapsules, the Folin-Ciocateau method was used as it allows for easy reference with a known standard, in this case, tannic acid [13]. Blank microcapsules were used to correct for the effect of wall material on the absorbance readings. Freeze-dried microparticles were evaluated by quantifying the amount of the bioactive compound on the surface of microparticles and the total amount of the bioactive compound loaded as described previously [8,14].

The in vitro release of tannin extract from the encapsulation matrix in the digestive system of ruminant animals was simulated using product solubility in various pH media following the procedure of Rossi et al [15]. Elution media used were: acetate buffer (pH 5.6), HCl buffer (pH 2.2) and phosphate buffer (pH 7.4) to simulate rumen, abomasal and intestinal conditions, respectively. In this study, the F57 ANKOM filter bags with a porosity of 25 μm used in fibre analysis was used. Microparticles (200 mg) were placed in the fibre filter bags and sealed using bag sealer, thereafter suspended in 50 mL of elution media, and agitated at 50 rpm inside an incubator at 39°C [11]. Aliquots of 1 mL were removed at 30 min, and at 1, 2, 4, 8, and 24 hours after incubation for analysis. The initial volume was maintained by the addition of fresh media. Aliquots taken were frozen immediately for subsequent analysis. The release of tannin was monitored by UV spectrophotometry as described above. The cumulative amount of tannin released at each time interval was corrected with the volume of the dissolution media. The data obtained for acetate was further fitted into zero order, first order, and Higuchi in vitro release kinetic models to find the best which predict the release of tannin extract in the rumen [16].

In vitro gas production study was performed following the procedure of Menke et al [17] with the modifications detailed in Adejoro and Hassen [18]. Rumen liquor was collected from two rumen-cannulated merino wethers fed lucerne hay (Medicago sativa) ad libitum, strained through four layers of cheesecloth into a pre-warmed thermos flask, and transported quickly to the laboratory. This rumen fluid was mixed with buffer-mineral solution and added to each serum bottle, which already contained approximately 400 mg of substrate, under a continuous stream of CO₂ and sealed with rubber stoppers. All bottles were placed inside an incubator at 39°C and 120 rpm. Using a semi-automated gas pressure transducer and digital tracker, gas pressure was recorded at 2, 4, 8, 12, and 24 h after incubation, and which was subsequently converted to volume. Gas samples were collected using a syringe and analyzed for methane concentration using gas chromatography (8610C BTU Gas analyser GC System, SRI Instruments, Bad Honnef, Germany). Eragrostis curvula hay (crude protein [CP], 55 g/kg; neutral detergent fibre [NDF], 784 g/kg; acid detergent fibre [ADF], 492 g/kg) and a total mixed ration (TMR) diet (CP, 180 g/kg; NDF, 301 g/kg; ADF, 214 g/kg) were used as substrates, and incubated with the crude ATE, or encapsulated ATE. The incubations containing an equivalent amount of the wall materials only (starch or maltodextrin/gum-arabic) were also included for comparison. For each substrate, treatments included i) diet only (control), ii) diet plus crude ATE, iii) diet plus starch-encapsulated ATE, iv) diet plus gum arabic-maltodextrin-encapsulated ATE, v) diet plus starch only, vi) diet plus gum Arabic-maltodextrin only. To each treatment containing ATE or encapsulated-ATE was added an equivalent of 25 mg ATE, which corresponds to 2.33% CT (leucocyanidin equivalent). Gas production data was fitted into the Ørskov and McDonald equation y = b(1–e^−ct) [19] to predict the rate and extent of fermentation (y = gas production at time t; b = slowly fermentable fraction, mL/g DM; and c = rate of fermentation of fraction ‘b’, mL/h). Rumen fluid pH was measured after 24 hours of incubation using a pH meter (Mettler Toledo 230 pH meter, Mettler-Toledo, Powai Mumbai, India) while ammonia-nitrogen concentration was analysed spectrophotometrically.

Results of microparticles sizes yield and EE were expressed as mean of at least four repeat batches. For the in vitro gas production, individual bottles within each run served as analytical replicates and were averaged prior to statistical analysis, while each run served as a statistical replicate. Gas volume was plotted against incubation time using the non-linear equation to predict fermentation kinetics [19]. Data on morphological parameters and gas production were analysed using the general linear model (GLM) procedure of SAS 9.3 (SAS Institute Inc., Cary, NC, USA) with the model:

Where, μ = overall mean, B_i = block effect (replicate), T_j = treatment effect, Y_ij = mean of individual observation and e_ij = residual error. Mean separation was done using Tukey’s test and significance was declared at p<0.05.

Scanning electron microscopy showed that the unencapsulated ATE particles were irregular in shape and had shiny or glossy surfaces with numerous impregnations (Figure 1). Their particle size ranged from 15 to 40 μm diameter, with most particles within the 20 to 25 μm range. Native starch (S-TE) and maltodextrin-gum arabic (MG-TE) encapsulated ATE ranged in size from 20 to 65 μm in diameter, with S-TE having more particles within 30 to 45 μm while MG-TE particles were mostly within 25 to 40 μm. The native S-TE microparticles were slightly bigger than the MG-TE microparticles. The S-TE microparticles were generally spherical or ovoid shaped, and dull in appearance while the MG-TE microparticles appeared in various flake shapes with mostly flat surfaces. There was a discernible effect of encapsulation on microparticle morphology, with the majority of them having a diameter of 25 to 40 μm, and the outer surfaces showing obvious modifications and evidence of adherence of the wall materials to the ATE particles in the form of several layers of coating.

In this study, the core (ATE) to wall material ratio was evaluated as the independent variable, capable of affecting the yield and EE of ATE. There were significant differences in surface and total tannin contents among the wall materials used, within each wall material (Table 1). The ratio of core to wall material also significantly influenced the surface and total tannin contents of microparticles. The ST contents of the microparticles ranged from 35.0% to 63.9% for S-TE and 8.43% to 11.2% for MG-TE. The total tannin recovered varied from 70.9% to 75.8% and 63.2% to 70.7% for S-TE and MG-TE microparticles, respectively (Table 1). The ST contents in MG-TE microparticles was generally lower than that in S-TE. Within each wall material, when the concentration of ATE was varied, total tannin recovered was not significantly different in both MG-TE and S-TE microparticles. The result of EE for MG-TE showed values of 64.8% at 25:75 (core to wall material ratio), 60.5% at 30:70 and 56.1% at 35:65. There was a decrease in EE with increasing concentration of ATE. For S-TE, the result showed that at 15% ATE concentration, EE was 48.8% while 33.4%, 27.7%, and 31.1% were recorded at 20%, 25%, and 30% ATE concentrations, respectively (Table 1).

The release profile of tannin from the unencapsulated ATE, S-TE, and MG-TE under optimal loading conditions, in acetate, phosphate and HCl buffer media at 39°C are shown in Figure 2a–c. In each dissolution media, a burst release pattern was observed within the first 4 hours of incubation for all microparticles across the different pH media. This phase was followed by a phase of gradual release from 4 to 8 hours after dissolution. Beyond, 8 hours, most of the tannin in the microparticles had been released into the dissolution media. For the unencapsulated ATE, about 75.5% of ATE was released within 2 hours and about 90.2% within 4 h after dissolution at 39°C. The release profile indicates that 58.7% and 75.8% of tannin was released from the MG-TE microparticles at 2 hours and 4 hours, respectively while 65% and 78.7% were released from S-TE after the same incubation times respectively. The in vitro release data obtained in acetate buffer, simulating rumen pH, showed that the release patterns of tannin from microparticles fitted into the Higuchi and first-order models better than the zero-order model (Table 2).

The ATE treatment reduced total gas production up to 24 h incubation when compared to the control diet in both substrates (p<0.0001). The inclusion of S-TE and MG-TE generally resulted in higher total gas production across the time intervals up to 24 h when compared with the control treatment in both Eragrostis curvula (EC) and TMR substrates. However, total gas production was lower in S-TE and MG-TE treatments than in incubations containing only the wall materials (Starch; gum arabic/maltodextrin) (p<0.0001). The ATE tended to reduce methane emission in both EC and TMR substrates (p<0.0001). The S-TE and MG-TE incubations did not reduce methane production when compared with the control or ATE treatment in both substrates (p<0.0001), however, in the TMR incubations, methane values obtained in S-TE were lower compared to the starch only treatment (p<0.0001). The intensity of methane produced when expressed as the ratio of methane volume to total gas volume showed that tannin inclusions did not affect methane concentration across the treatments in EC (p = 0.38) and TMR (p = 0.13) substrates. The rate of fermentation of the insoluble fraction with starch and S-TE was higher compared to the control, ATE and incubations with gum Arabic and maltodextrin in the Eragrostis hay substrate (p<0.0001). However, no difference in fermentation kinetics across the treatments was observed in the TMR substrate (p = 0.68). Supplementation with ATE, S-TE, or MG-TE did not affect rumen pH after 24 h incubation in both Eragrostis hay (p = 0.15) and TMR (p = 0.11) substrates. However, with Eragrostis hay as the substrate, ATE, S-TE, and MG-TE reduced ammonia nitrogen concentration in rumen fluid after 24 h in vitro incubation (p = 0.04) while no differences were observed in the TMR substrate (p = 0.11). The unencapsulated ATE was more effective in reducing total gas and CH₄ production than samples containing S-TE and MG-TE (Table 3).